search

AIZOOTech / FaceMaskDetection

开源人脸口罩检测模型和数据 Detect faces and determine whether people are wearing mask.
https://demo.aizoo.com/face-mask-detection.html
MIT License
2.08k stars 618 forks source link

Change resolution of video stream #36

Open grdsvt opened 4 years ago

grdsvt commented 4 years ago

Hi, is it possible to change resolution of video to 640x480px or it is necessary to rebuild the model? Thanks for support.

AIZOOTech commented 4 years ago

I did not do multi scale training. You can change the input to 640, remeber to change the 

feature_map_sizes = [[33, 33], [17, 17], [9, 9], [5, 5], [3, 3]]

to the corresponding size when input is 640x480. However， The result may not performs well. The best way is train the model with 640x480 input size.

grdsvt commented 4 years ago

Hi, changet the resolution to 640x480 and it is working fine. Do you suggest to re-train the model with this resolution? Whis is the resolution that you used? How have I change the feature_map_size?

Thanks for your availability

AIZOOTech commented 4 years ago

Yes, the best solution is to train the model with 640x480 input size.

grdsvt commented 4 years ago

I have problem with training......have you this model?

AIZOOTech commented 4 years ago

I can train it, however I am busy those days. Are you using this for commercial application? If so, we can train the new model for you if you can afford some payment.

grdsvt commented 4 years ago

no sir, I am studying deep learning to complete my course and I see that your work is very amazing. Thanks

AIZOOTech commented 4 years ago

Ok, this model is a simple SSD framework. Training process is quite easy. You can find many open source SSD implementation on Github. Hope everything is fine.

grdsvt commented 4 years ago

Thanks